Yeah, that's my understanding as well. But what I wonder is what would have happened if someone at the top had brought the C++ and .NET teams in and made some Jobsian declaration like "this isn't negotiable, the company is switching to .NET. All of your jobs depend on finding a way to make it work." Were those performance problems really insoluble, if all the engineering ability of the company were brought to bear on them? I get the impression that a lot of the C++ old hands were never pleased with .NET to begin with, and were all too happy to chuck it.
Right, it certainly sounded from the outside that a lot of the Longhorn performance issues were various sorts of sandbagging by one side or the other. There's an interesting leadership question there too if some of the worst sandbagging was intentional or just ADHD-style distractions due to a lack of a singular, focused vision for the product. (The biggest example being letting so much of the WinFS team get sucked into the never-ending "semantic web" rabbit hole of trying to "schematize the world"; it's hard to focus on performance when you are busy trying to catalog all the different ways that people store data.)
At RustConf 2017 keynote presented by Joe Duffy regarding Midori at a certain moment, he mentions that even with Midori running in front of them, WinDev guys were still not open to the idea of such kind of system being possible.
"Safe Systems Software and the Future of Computing"
